Appendix II 



Polarity Determination Under Uncertain 

 Outgroup Relationships 



I used a modified version of the outgroup method described by Maddison et al. (1984) and 

 M. J. Donoghue (pers. comm., 1982) to assess character polarities. This method assesses 

 the condition of the outgroup node (branch point linking the ingroup with its sister group 

 on a phylogenetic tree or a cladogram) in order to minimize character- state changes at all 

 hierarchical levels. Briefly, the cladogram for the ingroup and various outgroups is 

 rerooted at the outgroup node, and the conditions of the various subterminal nodes on the 

 rerooted cladogram are assessed using an optimization procedure similar to that of Farris 

 (1970). First, the terminal nodes (ends of branches) are labeled according to the condition 

 found in the outgroup occupying that position. Second, the subterminal nodes are assigned 

 character states according to the following rules: (1) If both nodes above the node in 

 question have the same state, assign that state to the node in question. (2) If the two nodes 

 above the node in question have different states, the assignment of the node in question is 

 equivocal (?). (3) If one node above the node in question is equivocal and the other is not, 

 assign the node in question the state of the unequivocal node. The state assigned to the 

 outgroup node (basal node of the rerooted cladogram) is taken as plesiomorphic. 



Because the relationships among iguanines and the outgroups used in this study are 

 poorly understood, I was forced to consider all possible cladograms for four unspecified 

 outgroups and an ingroup, of which there are nine (Fig. 60). After these cladograms are 

 rerooted at the outgroup node (Fig. 61), it can be seen that not all of them need to be 

 considered further, since many will yield identical assessments of the condition at the 

 outgroup node. Complete equivalence is seen between some of the rerooted cladograms: 

 A = E = G, and C = F. By swiveling branches about nodes, which does not alter the 

 relationships implied by the diagrams, rerooted cladograms A, B, and D are found to be 

 equivalent. Finally, given only the distribution of character states in the ingroup and these 

 four outgroups, the state assigned to the outgroup node in rerooted cladograms H and I 

 must be identical to that of the basal node in the clade formed by the four outgroups. 

 Therefore, for the purposes of this analysis, rerooted cladograms H and I can be 

 considered to be equivalent to A and C, respectively. Only two topologies need to be 

 considered further, A and C (Fig. 61). 



For any given character, the conditions of the outgroups can be placed on the terminal 

 branches of the two rerooted cladograms (Fig. 61 A, C) in all possible combinations, and 

 the condition of the outgroup node (i.e., the character's polarity) can be assessed. For 

 cases in which all four outgroups suggest a single interpretation, that interpretation is 

 accepted. For cases in which more than one of the states found in the ingroup also occur in 

 one or more outgroups, the polarity of the character is ambiguous. In such cases, I have 



179 



